How to Track Whether AI Engines Are Citing Your Brand

Your website ranks on page one of Google. Your reviews look solid. But when someone opens ChatGPT and asks "who handles [your service] in town," you have no idea whether your business comes up — or whether a competitor does instead.

That gap is the core challenge with tracking AI search visibility. The platforms shaping the most discovery queries in 2026 publish almost no data about what they surface. ChatGPT has no impressions dashboard. Perplexity sends no monthly report. Traditional rank trackers were never designed for conversational AI.

This guide covers what to measure, how to track it without an enterprise budget, and — critically — how to connect the data to actual business outcomes.

Why Traditional Analytics Miss the AI Picture

Google Analytics and Search Console measure clicks from links. AI-generated responses often give buyers the answer they need without sending a click at all — your brand can be cited as the recommended option and still show zero measurable traffic from that interaction.

Google has introduced AI Mode tracking in Search Console, which shows impressions and clicks from its dedicated AI Mode search experience. But this covers only one platform and one type of AI interaction. It tells you nothing about what ChatGPT says about you, what Perplexity surfaces, or how Gemini frames your competitors alongside your brand.

According to a February 2026 survey by Search Engine Journal, 67% of enterprise marketing teams now include AI visibility in their monitoring stack — up from under 20% two years earlier. The shift happened fast enough that brands relying solely on organic rank reports are increasingly working with an incomplete picture.

The deeper issue: ranking and citation are different things. Understanding what generative engine optimization means in practice helps explain why the gap exists — but tracking is how you confirm it in your own category.

The Three Metrics That Actually Matter

Share of Model (SoM)

Share of Model measures how often your brand is recommended across a defined set of prompts on a given AI platform. If you test 50 prompts and your brand appears as a recommendation in 12, your SoM is 24%. It is the closest AI equivalent to organic market share — and it is what actually drives discovery.

Citation Rate

Citation rate tracks how often a specific URL on your site appears in AI-generated responses, either as a linked source or as the basis for a summary. This is where technical factors — structured data, content clarity, internal linking structure — become directly measurable. A page earning citations on Perplexity but not ChatGPT usually signals a difference in how each platform weighs third-party endorsements versus on-page signals.

Share of Voice vs. Competitors

Share of voice puts your SoM in context. A 24% rate means something different if the next competitor is at 8% versus 41%. Most actionable tracking programs build a competitor prompt set alongside their own brand queries so both numbers are gathered in the same cycle.

One important caveat on all three: AI citations are unusually volatile. Research from Profound — which analyzes over 1.5 billion real user prompts — found that 40–60% of cited domains change month over month. Google AI Overviews shows a 59.3% monthly drift rate; ChatGPT sits at 54.1%; Perplexity at 40.5%. A single snapshot is noise. Trends over 60–90 days are signal.

Method 1: Manual Tracking With a Prompt Library

Manual tracking costs nothing but time. For most small businesses — a dental practice in Central Oregon, a specialty retailer in Bend, a home services company serving the region — it is the right starting point and often all you will need.

Build Your Prompt Set

Create 25–50 prompts organized across three intent types:

• Informational: "What should I look for in a [service category] provider?"
• Comparative: "What are the best [service category] options near Bend, OR?"
• Transactional: "Who handles [specific service] in Central Oregon?"

Include 5–10 prompts that name competitors directly: "How does [your brand] compare to [competitor]?" These reveal whether you appear in brand-adjacent conversations — typically a stronger purchase-intent signal than category queries.

Run Tests Consistently

Test across the platforms your customers actually use. For most local service businesses, the priority list is: ChatGPT, Perplexity, Google Gemini, and regular Google Search via incognito to catch AI Overviews. Add Microsoft Copilot if your audience skews toward Microsoft 365 users.

Running 25 prompts across 5 platforms totals 125 queries per cycle — roughly 3–4 hours if you are logging carefully. Run this bi-weekly and copy the full response text, not just a yes/no result. The wording of how your brand is described matters as much as whether it appears.

Your Tracking Spreadsheet

Log these columns for every test run:

• Date — essential for trend analysis
• Platform — ChatGPT, Perplexity, etc.
• Prompt used — exact text, unchanged between cycles
• Brand mentioned — Yes / No
• Mention type — Direct recommendation / Comparison / Category mention
• Position — First recommendation, second, buried in body text
• URL cited — which specific page, if a source link appeared
• Sentiment — Positive / Neutral / Negative / Mixed

After three bi-weekly cycles you will have enough data to identify your baseline SoM per platform and which prompt types produce the most revealing results. That baseline is what makes future changes meaningful — without it, you are reacting to noise.

Method 2: Google Search Console for AI Mode Data

Google introduced AI Mode tracking in Search Console, allowing publishers to see impressions and clicks from its dedicated AI Mode search experience. Open the Performance report, select "Search type," and look for the AI Mode option to separate this traffic from standard web results.

The coverage is limited. It reflects only Google's AI Mode interface — not AI Overviews embedded in standard results, and not ChatGPT, Perplexity, or standalone Gemini. Still, if Google drives most of your organic traffic, this data is useful for confirming whether AI-mode searches are absorbing impressions without returning clicks.

Pair this with UTM parameters on pages you expect are getting cited. If a service page earns significant AI Mode impressions but almost no clicks, that is evidence buyers are getting their answer without visiting — and a prompt to either optimize the page for click-through or accept the brand-awareness value of the mention.

Method 3: Dedicated AI Visibility Platforms

Manual tracking works until scale makes it impractical. When you have more than 50 prompts to monitor, need daily updates, or are in a competitive market where citation rankings shift weekly, purpose-built platforms become worth the investment.

The main options in 2026:

• Profound — Enterprise-grade, tracks across 10+ AI engines including ChatGPT, Gemini, Perplexity, Meta AI, and DeepSeek. Raised $96M in February 2026. Best for brands spending $5k+/month on content and needing competitive intelligence at scale.
• Semrush — The AI Visibility module in Semrush One adds brand citation tracking on top of its existing SEO suite. Lower learning curve for teams already using Semrush; less depth than a purpose-built tool.
• LLMrefs — Keyword-style rank tracking adapted for AI search. Well-suited for SEO-led teams who want AI data in a format that mirrors how they already report.
• Peec AI — Daily monitoring across ChatGPT, Perplexity, and Google AI Overviews. Mid-market pricing, well-suited for agencies managing multiple clients.

Starter packages covering 100–200 prompts per month typically run $150–$400/month. Enterprise tiers with unlimited prompts and full competitor tracking start around $1,000–$2,500/month.

How to Interpret What You Find

Not all citations carry equal weight. Appearing fifth in a list buried in the fourth paragraph of a ChatGPT response is different from being named first in a Perplexity summary with a direct source link to your service page. Your tracking log should capture position and context, not just presence.

Four things worth examining closely:

• Position consistency — Are you reliably first, second, or third across similar prompts? High volatility here suggests AI engines lack strong signals about your authority relative to competitors.
• URL diversity — Which pages get cited? If citations cluster on your homepage but not your service or FAQ pages, those pages likely need better structure for AI retrieval. See how to optimize your content for LLM citations for specific fixes.
• Platform divergence — Consistently appearing in Perplexity but not ChatGPT usually indicates a gap in third-party web presence. ChatGPT weights broader web authority — press, directories, reviews; Perplexity leans more heavily on real-time retrieval.
• Competitor drift — If a competitor's SoM rises 10 points in one cycle, check what they published or where they earned coverage in the prior 30 days. That often reveals exactly what signal the platform rewarded.

The revenue connection is harder to measure directly. AI platforms do not pass UTM parameters the way a paid ad does. The most reliable signal is a correlation between improved citation tracking and an uplift in branded direct traffic or branded search queries. The ROI calculator walks through how to frame what that visibility shift is worth in dollars.

When This Is NOT the Right Solution

You do not have a content strategy to act on the data. Tracking reveals gaps; gaps only close through content improvements. If your team cannot realistically update existing pages or publish new material based on what the data shows, the tracking cycle produces information without any ability to respond. Build the editorial capacity before the monitoring.

Your customers do not discover you through AI search. A trades business where 80% of leads come from referrals or direct calls faces no meaningful AI citation risk. Neither does a professional service firm where the buyer relationship starts from a specific referral rather than a discovery query. Audit the actual source of your leads before adding a new analytics layer.

Your market is entirely hyperlocal. AI engines are significantly less reliable than Google Maps at surfacing local service businesses for geo-specific queries. A contractor serving only the Bend, OR metro area will see more return from a strong Google Business Profile and a consistent review strategy than from AI citation monitoring.

You have not completed traditional SEO basics. AI platforms largely derive their citations from the same high-authority signals as conventional search: quality backlinks, consistent business information, clear page structure, and third-party mentions. A brand that has not built those foundations will see more return from fixing them than from AI-specific tracking work.

Getting Started This Week

The fastest path to a real baseline: run 25 prompts across ChatGPT and Perplexity today, log the results in a spreadsheet, and repeat in two weeks. You will have a first trend line within a month — no tool spend required.

If you want help building a prompt library tailored to your service category, or want to work through whether the manual approach makes sense to scale to a paid platform, book a demo and we can review the data from your specific market together.

Frequently asked questions